For example, the diploid perennial species of Microseris and Agoseris (Western North American composite species related to the dandelion) occupy cooler and more mesic (having a moderate water supply) habitats than do the annual species that are adapted to warmer, more xeric (having a low or inadequate water supply), time-limited habitats. The annuals possess genomes that are only 30-60% the size of those of the perennials.

A similar pattern is observed among populations of the annual M. douglasii, in which DNA amount varies over 20%. The higher DNA values are restricted to plants growing in more mesic sites, generally on well-­developed soils. Sites with lower rainfall and/or poorly developed soils dry out more quickly than mesic sites and reduce the time available to complete the plant's life cycle. In these species, DNA amount may be subject to natural selection through its nucleotypic effects on cell size, mitotic cycle time, and the rate of development.

Repetitive DNA and plant genome evolution
The dynamics of genome size evolution are speculative. In current evolutionary theory, genetic differences among related species are viewed as resulting from evolutionary forces, for example, selection and/or genetic drift, acting on existing variability within populations. However, an analogous hypothesis maintaining that accumulation of small deletions or duplications is the primary factor in changing genome size is not supported by the distribution of DNA contents.

Genome size commonly ranges severalfold among congeneric species, but well-documented examples of intraspecific variation in genome size are rare; genome size is typically very constant within a species. Since there is apparently a lack of existing variation in DNA content within species on which evolutionary forces can act, a hypothesis invoking events that generate larger discontinuous quantitative changes seems more attractive. Such events, if occurring, may be relatively rare and triggered by hybridization, environmental stress, or genetic events that destabilize the genome.

For the most part, variation in genome size is not caused by reiteration of coding genes, but is due to differences in the amount of repetitive DNA. Repetition of sequences, tandemly arranged and/or dispersed throughout the genome, comprise the bulk of eukaryotic DNA. Clusters of tandemly repeated sequences consisting of short repeats as small as a few basepairs long (microsatellites or simple sequence repeats) are dispersed throughout the genome. Longer tandemly repeated sequences, typically about 150 bp to 400 bp, are often of1 around centromeres of chromosomes.

Genome growth via retrotransposon accumulation
A mechanism to generate dispersed repetitive DNA sequences involves the movement and accumulation of retrotransposons, mobile DNA sequences that can move from one genomic location to another by producing ribonucleic acid (RNA) I :s transcribed by reverse transcriptase into DNA that is t. inserted at a new site. Members of a retrotransposon family may be several kilobasepairs long and have copy numbers up to several hundred thousand. The large copy numbers t can be attained by retrotransposons indicate that their amplification is one of the forces leading to the growth of a genome.

Genome reduction via retrotransposon crossing-over
Evidence is accumulating that indicates reductions in genome size have been common evolutionary events. Recombination in homologous regions of loops resulting from pairing of retrotransposons located in the same chromatid may be one mechanism for reducing the size of genomes that he extensively accumulated retrotransposons (Fig. 2). Although: there is evidence from DNA sequencing that such  recombination events involving retrotransposons may occur in plants, it is not known if they occur frequently enough facilitate rapid evolutionary reductions in genome size.
Genome size and angiosperm phylogeny

It has been proposed that plants may have a "one-way tic to genomic obesity" as a consequence of retrotransposon accumulation and polyploidy (having one or more extra sets of chromosomes). Initial evaluations of the genomic obesity hypothesis have recently been conducted by superimposing estimated ancestral genome sizes onto well-established phylogenies (branching diagrams of evolutionary history),1 cotton (Gossypium) and its closely related genera in the cotton tribe (Gossypieae) and of the angiosperms to determine whether genome size only increases over the course of evolution or whether both increases and decreases occur.

In the cotton tribe, DNA content among diploid species varies over about a sevenfold range. Based on comparison of well-supported phylogeny for cotton and its allies with I pattern of evolutionary history generated from an analysis of

Fig. DNA detection by intrachromosomal recombination between (a) paired homologous terminal repeats of a single retrotransposon and (b) paired retrotransposons located on the same chromosome.

DNA content of living species and estimated ancestral genome sizes, it was inferred that both increases and decreases genome size occurred during the evolution of this tribe. It found that the frequency of reductions actually exceeds that of increases.
Recent advances in plant taxonomy by participants of I Angiosperm Phylogeny Group using molecular phylogenetic data, including DNA sequences, along with nonmolecular phylogenetic data, have resulted in a revised classification the angiosperms that reflects evolutionary relationships. This phylogenetic framework and the availability of a list (currently 3500) of angiosperm genome sizes in the Plant DNA C-value Database are leading to analyses of evolution of genome size in relationship to angiosperm evolution. From initial broadly based analyses, it appears that the ancestral genome size for angiosperms was relatively small (1.4 pg), with the trend for the growth of genomes during the divergence of angiosperms.

However, apparent decreases in genome size accompanying angiosperm phylogeny were also detected. Large genomes (35 pg) in monocots and eudicots are confined to taxa occupying derived positions within larger clades. As additional genome sizes are added to the Plant DNA C-value Database, a more detailed picture of angiosperm genome size evolution will certainly be produced. It appears that a dynamic model that includes both increases and decreases for genome size evolution in plants is emerging.